The definitional consequences of duties of probability being probable consequences, it follows that that same probability is the test of their proximateness. The definitional consequences of such a duty, when they actually happen, are necessarily consequences that were probable and therefore proximate; and no other consequences are proximate to the act, except further consequences of those definitional consequences, the proximateness of which will be discussed below.

To the principles just stated there is perhaps an exception. When a negligent act consists in setting in operation an active dangerous agent, and is a breach of duty because that agent will probably do harm of a certain kind to persons or things within the range of its activity, such harm to such persons or things being the definitional consequence of the duty in specie, it has been held that if the activity of the agent extends further than was probable, and causes injury of such a kind to a person or thing outside of the probable field of its activity, so that such injury was in fact improbable and not a definitional consequence of the duty in the particular case, such injury must nevertheless be treated as standing on the footing of a definitional consequence. The duty must be deemed to have corresponded to the right so violated, and the injury to be proximate to the act.

A railroad company negligently set fire to some dry hedge clippings on its own land. The fire ran five hundred yards across a stubble field to the plaintiff's house and burned it. A verdict for the plaintiff was sustained.21 Some of the judges thought that the injury to the house was probable. On that view the case presents no difficulty. One judge dissented on the ground that the injury was not probable. But several of the judges said that the company would be liable even though it was not probable that the fire would run so far, if the company had been negligent in starting the fire. The company, of course, had a right to set fire to the clippings, which were its own property, and might have done so intentionally. Negligence here must therefore have meant that the act was unreasonably dangerous because of the probability that the fire would spread to neighboring land. The duty not to set fire to the clippings, a duty of probability, was undoubtedly owed to the owners of all land

21 Smith v. London & S. W. Ry. Co., L. R. 5 C. P. 98, 39 L. J. C. P. 68, L. R. 6 C. P. 14, 40 L. J. C. P. 21 (1870).

to which the fire would probably spread, and corresponded to their rights. Injury to them would undoubtedly have been proximate. But a recovery by the plaintiff, if the injury to him was not probable, could be supported only on the ground of an exception to the general rule such as is now under consideration. The language of the judges, however, suggests that they did not distinguish between the definitional consequences of the duty and further consequences of them, which further consequences some courts, as will presently be explained, have held to be proximate though not probable. A similar decision was made in a case where the defendant shot and wounded a dog, and the infuriated animal did damage outside of the probable area of his activity; and the court said explicitly that a person who sets a dangerous thing in action is liable even for improbable consequences.

22

When a building has been negligently set on fire and the fire has spread to neighboring buildings, some courts have held the damage to the latter remote on grounds of humanity, saying that to hold a negligent person liable for a whole conflagration would be to impose a too crushing liability.23 Such decisions have no bearing on the general theory of proximateness. Other courts have refused to follow any such principle, and various distinctions have been drawn. On the whole the cases respecting the spread of fire must perhaps be regarded as sui generis affording little guidance for other kinds of cases.

In duties of intention all the definitional consequences of the duty are intended. Therefore, if any such consequences happen, they are proximate. The same rule applies here as to duties of probability, that to make a tort some definitional consequence of the duty must actually happen, and all further injurious consequences, to be proximate, must be consequences of that definitional consequence. Of course the same act, being done with an intention to injure one person and being therefore a breach of a duty of intention toward him, may be negligent as toward another person, and a breach of a duty of probability as to him. In that case if the latter person is in fact injured, the proximateness of the injury to him may depend upon probability, not upon intention.

2 Isham v. Dow's Estate, 70 Vt. 588, 41 Atl. 585 (1898).

"Ryan v. New York Central R. R. Co., 35 N. Y. 210 (1866); Pennsylvania R. R. Co. v. Kerr, 62 Pa. St. 353 (1870).

If the definitional consequences of the duty broken, having become actual as aforesaid, are themselves such consequences as constitute the violation of right necessary to make a tort, which may be called violative consequences, of course the same tests of proximateness apply to them in one aspect as in the other. If A. fires off a gun when that is a negligent act, i. e., is unreasonably dangerous, because of the probability that he will hit B., and B. is hit, the hitting of B. is both the definitional consequence of the duty and the violative consequence. In either aspect it is proximate because it was probable.

But often the violative consequences are not identical with the definitional ones, but follow them as further consequences of the act. If so, in order that they may be proximate consequences of the act, it is necessary, and is sufficient, that they be actually and proximately consequences of the definitional consequences. In considering the proximateness of such a violative consequence, the definitional consequence, which more directly produced it, is taken as its cause rather than the act which was its primal cause. The question is: assuming that the definitional consequence actually happens, is such a violative consequence proximate to it? It follows that, if in any case probability is to be taken as the test of the proximateness of a violative consequence, the probability of the violative consequence following the definitional consequence is not to be compounded with the probability that the definitional consequence would follow the act. For example: If the probability of the definitional consequence following the act was 12 and the probability of the violative consequence following the definitional one was also 12, then the probability of the violative consequence being produced by the act was at the outset only 1⁄2 x 1⁄2, or 4. But in estimating the probability of the violative consequence the definitional one should be taken as certain, i. e., to be represented by unity, the true question being, will the definitional consequence, if it happens, produce the violative one. The probability of the violative consequence therefore should not be taken as 14, but as 1 x 2, or 2. There is very little direct authority for the above rule against compounding probabilities. In fact, numerical calculations of probabilities are seldom made, and generally would not be worth much for practical purposes. However, there are cases

where the principle of not compounding the probabilities of suc

cessive or coöperating causes seems to have been approved. The courts sometimes say that the final consequence of a series, which is held proximate as being probable, was not probable at the outset, that is, would not have been probable if all the probabilities had been compounded, but is probable on the assumption that some intermediate consequence actually happens.24

The question whether a violative consequence was proximate to a preceding definitional consequence which was its more immediate cause, arises in two classes of cases: (1) where the definitional consequence was itself defined relatively, by reference to the violative consequence, and (2) where the definitional consequence was defined absolutely, not by any such reference.

A definitional consequence of a duty may itself be defined as consisting in the existence of some condition of things that in its turn will or may produce a violative consequence, such a causal relation to a violative consequence being of the essence of its nature as a definitional consequence of the duty. This is often the case in duties respecting dangerous things. The mere existence of a dangerous thing may be the definitional consequence of the duty, so that if the dangerous thing be produced, there will be a breach of the duty even though it does no harm. But a dangerous thing means a thing that will probably do harm. The harm, which in case of a tort will be the violative consequence, is a consequence of the existence of the dangerous thing, and is that by reference to which the existence of the dangerous thing is defined as being the definitional consequence of the duty. Thus in a duty to keep a highway safe, the condition of the highway itself, and no more than that, is the definitional consequence of the duty. The duty is not properly defined as a duty not to cause harm to travellers by the highway being unsafe, but simply as a duty to keep it safe. If the highway is allowed to remain in an unsafe condition, the duty is broken. But the probability of harm to travellers, which would be a violation of the right to which the duty corresponds, is what makes a given condition of the highway an unsafe condition.

When the definitional consequence is defined relatively by reference to a probable violative consequence, the same principles apply between those two consequences as between the act and the defini

24 Quigley v. Delaware & H. Canal Co., 142 Pa. St. 388, 21 Atl. 827 (1891).

tional consequence in a duty of probability. To make a tort, the violative consequence must actually happen, and in order to be proximate it must be probable. The test of proximateness here is probability. If, for instance, a traveller is hurt because of the dangerous condition of the highway, that injury must have been a probable consequence of the highway being in such a condition.

But a definitional consequence may be defined by reference not to its probable, but to its actual consequences, though such cases are rare. If, for instance, A. maintains some thing on his land which sends noxious gases onto B.'s land. The thing is a nuisance; and it is so even though the emission of such gases or their passing onto B.'s land was improbable, i. e., if A. had taken such precautions as would probably prevent the emission of any gases. The duty in such a case, I think, should be defined as a duty not to maintain such a thing, not as a duty not to send gases onto B.'s land. The existence of the thing would be the definitional consequence of the duty, and the passing of the gases onto B.'s land the resultant violative consequence. But the thing would be defined as a thing that would produce such a result. In this case, I believe that the test of proximateness should be the same as for the definitional consequences of peremptory duties, not probability but isolation. If gases actually passed onto B.'s land and did damage there, I think that the damage should not be held remote because it was not probable. In fact, some courts have applied the test of isolation and some of probability.

The defendant bought from the plaintiff a narrow strip of land along a stream, and then built a dam on his own land below. He made an embankment on the strip to protect the plaintiff's land from being flooded, but the water percolated through it and made the plaintiff's land wet. The defendant was held liable, though he had built the embankment with due care and skill, i. e., in such a manner that the percolation was not probable. The court said that the percolation was not due to the act of God, i. e., there was no isolating cause, and was a proximate consequence of building the dam.25 But where the defendants made a cesspool on their own land, from which filth percolated into the plaintiff's land, it was held that

25 Pixley v. Clark, 35 N. Y. 520 (1866).